Tracked down two bugs to get the right proximity into the hold copy map

Tracked down two bugs to get the right proximity into the hold copy map

The first took a little while because I was learning the lay of the
land and needed Mike's help.  SELECTs in
action.hold_copy_calculated_proximity() didn't quite filter their
result rows hard enough.

The second was more subtle.  After fixing the first bug, all proximities
reaching the copy map after targeting were zero.  This came down to one
place where we mistook a parameter that really means 'copy location' to
mean 'hold pickup lib', and wound up measuring the length of a path
whose two points were both hold pickup lib (so always zero).  Finally
tracked down and repaired.

Signed-off-by: Lebbeous Fogle-Weekley <lebbeous@esilibrary.com>